All errors returned from NetProbe are managed as fatal later.
Decide, connection issues are not fatal but bad configuration is.
Without this configuration errors are silently ignored here.
Windows doesn't seem to like this address.
Also default to the fallback resolver IP if there is one and
no netprobe_address option in the configuration file.
Fix netprobe_timeout = -1 by the way
Write at least 1 byte. This ensures that sockets are ready to use for writing.
Windows specific: during the system startup, sockets can be created but the underlying buffers may not be setup yet. If this is the case Write fails with WSAENOBUFS: "An operation on a socket could not be performed because the system lacked sufficient buffer space or because a queue was full
This fixes: https://github.com/jedisct1/dnscrypt-proxy/issues/841
This works over DNSCrypt and DoH, but requires a specifically configured
server.
Instead of sending the actual DNS queries, the SH-T system works as follows:
Step 1: the client query is evaluated through Argon2id, a military-grade,
memory-hard, CPU-hard stretching function. This makes it very expensive
for an attacker to find the original query, even using GPUs and ASICs.
For post-quantum resistance, we use it to generate a 1024-bit key.
Step 2: in case the Argon2id algorithm has a vulnerability, or, since this
is a popular function used for hashing passwords and for cryptocurrencices,
and people may have built rainbow tables already, we use a hash function over
the result of the previous function. This immediately defeats rainbow tables.
Step 3: the output of the hash function is truncated to 64-bit.
Due to a property of this operation known as collision-misresistance, and even
if the previous steps fail due to a nation-state actor, it is impossible for a
server operator to prove what exact query was originally sent by a client.
This feature is experimental.
in case we want to add user_group and whatnot.
Remove the command-line option as it hides the caveats documented
in the configuration file.
Remove TODO. TODO statements always remain in that state forever.
This currently doesn't replace the previous process. Maybe there is a way to achieve
this in Go. Need to look closer at os.exec
Also start-child -> child